By default when rsync transferrs a file named foo it will create a temporary file named .foo-xxxx and then rename it over the previous file. The advantage of this is that there is an atomic operation replacing one version of the file with another. The disadvantage is that if you use a filesystem like BTRFS or ZFS with deduplication disabled (as you will do unless you have unusually large amounts of RAM to space) then creating a new file will break copy-on- write.
I have some situations where I want to transfer backups of data to ZFS and BTRFS filesystems and not use excessive disk space by using multiple copies of database files. I'm aware of send/recv functionality, but it's not ready for use on BTRFS and on ZFS it requires that the sender be running ZFS. Can I make rsync overwrite files in place instead of recreating them? If so how? Yes I've read the rsync man page and didn't find it. No I didn't read through all 3446 lines (50+ A4 pages) of the man page, a search didn't find it. -- My Main Blog http://etbe.coker.com.au/ My Documents Blog http://doc.coker.com.au/ _______________________________________________ luv-main mailing list [email protected] http://lists.luv.asn.au/listinfo/luv-main
